Automated Podcast Transcription

Programmatically transcribe audio streams and surface the well-known entities contained therein.

Installation requirements

• sbt 1.x
• A local installation of Node.js and npm
• A local installation of SoX
• A Google Could Platform (GCP) project with the Cloud Speech-to-Text and Natural Language APIs enabled

Usage

Ensure that your local environment is targeting the desired GCP project:

$ export GOOGLE_APPLICATION_CREDENTIALS=/path/to/your/project/credentials.json

From the project root, begin an interactive sbt session:

$ sbt

From the interactive sbt session, issue:

> project gui
> run --port 8080

Front-end/GUI (node/npm) dependencies will be downloaded via sbt.

Visit http://localhost:8080 in a browser to begin processing an audio stream.

Project structure

common

Shared resources (domain objects, [de]serializers, etc.) used by both the api and gui projects.

api

A REST API for processing audio streams and performing the subsequent entity detection.

gui

A web application and server, with browser-based GUI.